diff --git a/comfy/ldm/wan/model.py b/comfy/ldm/wan/model.py
index 63ab3fbfc..1d6edb354 100644
--- a/comfy/ldm/wan/model.py
+++ b/comfy/ldm/wan/model.py
@@ -440,7 +440,7 @@ class WanModel(torch.nn.Module):
 
         # embeddings
         self.patch_embedding = operations.Conv3d(
-            20, dim, kernel_size=patch_size, stride=patch_size, device=operation_settings.get("device"), dtype=torch.float32)
+            in_dim, dim, kernel_size=patch_size, stride=patch_size, device=operation_settings.get("device"), dtype=torch.float32)
         self.text_embedding = nn.Sequential(
             operations.Linear(text_dim, dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype")), nn.GELU(approximate='tanh'),
             operations.Linear(dim, dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype")))
diff --git a/comfy_extras/nodes_wan.py b/comfy_extras/nodes_wan.py
index 3563b3ad7..d929eee39 100644
--- a/comfy_extras/nodes_wan.py
+++ b/comfy_extras/nodes_wan.py
@@ -672,11 +672,11 @@ class WanTrackToVideo:
                 mask, video = motion_patched[:, 0:4], motion_patched[:, 4:]
                 # Add motion features to conditioning
                 positive = node_helpers.conditioning_set_values(positive, 
-                                                                {"concat_mask": mask,
-                                                                "concat_latent_image": video})
+                                                                {"concat_mask": motion_patched,
+                                                                "concat_latent_image": image})
                 negative = node_helpers.conditioning_set_values(negative, 
-                                                                {"concat_mask": mask,
-                                                                "concat_latent_image": video})
+                                                                {"concat_mask": motion_patched,
+                                                                "concat_latent_image": image})
                 
 
         # Handle clip vision output if provided